DLSS5 looks bad, but after using for a little while frame generation does have some use cases. It requires a high refresh monitor to be worth anything first and foremost to ensure there’s space for the interpolated frames between a high base frame rate. I use a 144hz monitor, so with 2x frame rate I still get a perfectly playable 72fps based latency.
I found it to be good while playing Kingdom Come Deliverance 2 because it let me turn up the other settings while consistently looking smooth. It’s a slow enough game that the increased latency isn’t noticeable.
It’s especially helpful for CPU bound games like Helldivers 2 and Victoria 3 because the interpolation happens after the main rendering. I could not get HD2 above ~90fps with my CPU, but with frame gen on the CPU only had to render 72fps. It also meant I could turn DLSS scaling off, which meant the generated frames were based on native renders and everything looked better. Victoria 3 went from a painful 20-30fps to a much more tolerable 4-60.
All that said it does have serious issues. The increased latency means I’ll never use it for a reaction time based shooter like Enlisted or Deadlock. A few games just don’t look good with it. Star Wars Jedi Survivor had poor performance anyway, but turning frame gen on both makes it feel like it plays worse while also adding terrible ghosting effects and artifacts.
I’ve been interpolating movies and tv shows since 2015, so I think I was better able to set my expectations for the downsides of interpolation.
Okay, thank you for describing that, I drew a couple charts to help me think through the situation. The lettered frames represent the image sitting on the gpu. It’s called B1 because it’s a frame that needs to be interpolated from frame B, so frame B needs to be rendered first before interpolation can occur. The numbers represent units of time, so in the first chart at time unit 2 Frame A is being displayed on the screen while Frame B is being rendered by the card.
The first chart I drew I made an assumption that the interpolation is basically free due to my understanding that the DLSS functions use specialized cores that could be running in tandem with the rendering of frame B. Then I realized that was an assumption and went over it again assuming frame B needed to be fully rendered and so interpolation would need to take a unit of time before displaying.
I think this illustrates what you’re saying.